August 8, 2024
How AI Will Change the Risk Landscape
A new report from Swiss Re (“Tech-tonic Shifts: How AI could change industry risk landscapes”) studies the risk posed by AI to ten major industries. It also identifies the opportunities for insurers to help clients mitigate these risks.
One estimate suggests that generative AI, a sub-branch of AI, could alone add “between USD 2.6 trillion and USD 4.4 trillion” annually to the global economy. However, the report points out, AI is no different from other technologies: it can go wrong.
It may fail against performance benchmarks; it may inadvertently perpetuate discrimination; it could be subject to malicious attack; or it will perhaps cause real world damages. Wherever there are opportunities – and the opportunities are huge – there will be risk.
The report ranks ten major industries in terms of exposure to AI risk both immediately and over the next ten years. Drawing on data and anecdotal examples, these are sectors where AI risk is currently most concentrated, both by frequency and severity; and where it may become more prevalent with time. Here are some highlights:
Technology: This sector should surprise no one given its role as the developer of AI systems.
Energy and Utilities: Risk here should be relatively low; even though any incident is likely to have a high severity. Frequency will increase as AI becomes more widely deployed in smart grids.
Media and communications: There is short-term exposure to intellectual property risk, reflecting the recent coming onstream of generative AI and the legal status of the use of copyrighted materials to train large language models (LLMs). As AI use becomes more common, the risks will be more disbursed over sectors.
Healthcare and Pharmaceuticals: AI risk to these industries is expected to be prominent due to a combination of 1) high frequency of potential incidents, given a high number of applications across the health value chain that could use AI; and 2) potentially high-severity losses (e.g., bodily injury, professional liability).
Here is how Swiss Re ranks AI risk exposure by industry right now and in the near future (estimated rank in ten years appears in parentheses).
- (4) IT services
- (3) Energy and utilities
- (1) Health and pharma
- (10) Other services (retail, hospitality, legal)
- (2) Mobility and transportation
- (7) Financial and insurance services
- (6) Government and education
- (8) Manufacturing
- (5) Media and communications
- (9) Agriculture, food and beverages
Long Term Outlook
Overall, risks relating to ethics, bias and privacy will be more prominent in the short term. As AI models are established, the danger is that existing flaws in data storage and analysis will become entrenched. Industries holding sensitive personal data — such as healthcare, finance or law — are particularly vulnerable. Longer term, performance risk will grow in importance. Assuming teething pains in creating AI models can be solved, performance risk will become the dominant risk category. This will be the case especially for closed-systems production, such as agriculture or manufacturing. You can download a copy of AI and the industry risk landscape by Swiss Re here: https://tinyurl.com/undj6wxp
To mitigate these challenges, insurance companies should prioritize ethical AI development, leverage diverse and representative training data, evaluate and audit their AI systems on a consistent basis through a robust governance model, and maintain transparency in decision-making.